A Concise Overview of Recent Advances in Chatbot Technologies

By Precy Kwan, Grakn Labs.

Editor's note: Originally posted as part of Grakn's Advent at Grakn Labs series. #GraknLovesTech

2016 has been an interesting year for chat bots with both tech giants and startups taking a crack in the space. In this era, one can safely say that creating your own chat bot is no longer reserved only for crazy engineers but even non-technical people can create their own. Now on to the news!
Chat bots have received a lot of hype in 2016 with large news networks like CNET lauding it as the best thing since sliced bread.

So did 2016 actually end up being the year of the bot?

One of the pioneers of the rise of the bots is Chinese mobile messaging platform WeChat who has provided a bot platform since 2013. The bot platform has been incredibly successful in China where WeChat has more than 700 million users. People has been using it as part of their customer service offerings and as a way to notify their subscribers/customers about news. Though not strictly from this year, one WeChat bot celebrity is Microsoft’s Xiaoice whose natural language processing capabilities and entertainment value has elevated her to be a strong candidate as your next BFF.

So how have others responded to WeChat’s success?

Earlier this year in April, Facebook Messenger launched their chatbot platform for businesses in response to competition from other chat platforms like Kik and Telegram who already have their own bot platforms. Facebook offers developers the tools to “create a bot in 10 minutes with Node.js”. This includes an API to send/receive messages and to configure the sequencing of your bot’s response. Unfortunately, despite its high profile launch, performance left something to be desired. Facebook has admitted since the launch that the functionality provided in the beginning isn’t the most attractive but the tech has grown and matured over the year. Regardless, many high profile companies across various verticals have launched their own Facebook messenger bots from the Guardian, Disney and Healthtap. Well here’s the video straight from the horse’s mouth.

Microsoft is another contender in the bot arena, providing their Bot Framework to developers in March of this year. It allows developers to create bots to interact via text, Skype, Slack, Facebook Messenger and (obviously) its enterprise chat applications Teams and Office 365. They provide a SDK in C# and Node.js for developers to start building on top. In another classic Microsoft move, you can also develop your bot in its cloud service Azure.

Perhaps an overzealous attempt to show off its chatting capabilities, Microsoft created Tay, a Twitterbot that ignited a maelstrom of contempt with its sexist, racist and drug-abusing tweets. Originally, created to connect with millenials and the mercurial teenage population by mimicking their speech patterns, Tay’s ML algorithms went “totes cray” and started spamming its followers and was forced to be taken offline.

Kik, IBM Watson amongst others have all joined the bot bandwagon. There are also a slew of startups that created platforms for developers, non-developers and everyone in between. So the competition is really heating up.

Botsify is one of these startups that provides a graphical UI that requires no coding to get you started. It has landed some high profile clients like Travelex, Universal Music, Apple and Shazam. Rebot.me is another startup with some grand ambitions to bring bots to the average non-engineer Joe (or Jane). It works by having you “teach” your bot to respond to certain questions, so the more you talk to it the “smarter” your bot becomes. Not surprisingly, these bots are probably the most rudimentary in their functionality but still fun. Further along the scale, Motion.ai provides an API for developers as well as a pretty GUI to help the process along. It’s a bit more sophisticated than the “no-coding” options but more accessible to rookies with its GUI. For all of these startup options, the pricing is quite reasonable so it is accessible even for SMEs. Incubated by Disney and Techstars, Imperson is another bot provider that promises a better NLP experience and capable of showcasing brand identity. However, it doesn’t see, to be available to the public yet.

In summary, creating your own bot has never been easier or more affordable than it is now with a variety of developer tools available from different platforms. However, these bots are still relatively rudimentary and are capable of only simple commands and recognizing keywords. I think that there’s still some ways to go before accessible bots are capable of providing a more natural interface for users to interact with. I am generally bullish on bots and their ability to change the way businesses interact with their consumer base. Keen to see what additional functionalities will be brought to market in the upcoming year!

In typical #GraknLovesTech fashion, we end on a on-topic comedy video. Enjoy!

Precy Kwan is Chief Business Officer at Grakn Labs. Offshore bean counter turned microfinancier turned tech head. Masters in Land Economy and RTPI Prize from the University of Cambridge. You can find Precy lurking at coffee shops, ramen joints and pub quizzes around London.

GRAKN.AI is an open source distributed knowledge graph platform to power the next generation of intelligent applications. Using the power of machine reasoning, we provided a platform to help manage and make sense of highly interconnected big data. Grakn performs machine reasoning through Graql, a graph query language capable of reasoning and graph analytics.

Original. Reposted with permission.

Related:

• The Human Vector: Incorporate Speaker Embeddings to Make Your Bot More Powerful
• A Concise Overview of Recent Advances in Vehicle Technologies
• Deep Learning for Chatbots, Part 1 – Introduction